One endpoint, infinite images.

Create a token in your dashboard under API tokens. Send it on every request. The same token works for the REST API and the MCP server, and identifies which account is billed.

curl https://render.pixeldrive.pro/v1/render \
  -H "Authorization: Bearer pdrv_your_token_here" \
  -H "Content-Type: application/json" \
  -d '{ ... }'

A missing, revoked, or expired token returns 401.

Design a template once, then make any layer dynamic: select it and click the database icon to mark it a Variable. Each variable becomes a key in your render payload, addressed by its layer name (set one in the Layers panel) or its element id. Every template ships a ready-to-paste payload with all its variables on its API page — open a template → API.

Variable types

Text that never breaks the design

Long values auto-shrink to fit their box, so a 5-word headline and a 25-word one both stay inside the layout — no overflow, no clipping. Need control per render? Two optional overrides on any text variable:

{ "headline": { "text": "Big news", "fontSize": 64, "color": "#E11D48" } }

Translate (any language)

Add lang to a text value and the render service translates it into that language before drawing — self-hosted, cached per text + lang, and auto-fit so longer translations stay on-layout. Use an ISO-639-1 code (es, fr, ja, ar, …) — 70+ common languages, or any FLORES-200 code (spa_Latn) for the full 200.

{ "headline": { "text": "Spring Sale", "lang": "es" } }   →   "Venta de Primavera"

Or translate the whole template at once: pass a top-level langand every text layer renders in that language — no per-field setup, no need to re-send the copy. On the URL endpoint it’s just ?lang=es.

POST /v1/render   { "templateId": "…", "lang": "es" }
GET  /v1/render?templateId=…&token=…&lang=es     # whole template, in Spanish

Fonts

Whatever typeface you set in the editor — web fonts or uploaded custom fonts — is embedded in the template and rendered pixel-for-pixel on the server. No per-request font config needed.

Output formats

Default output is PNG at the template’s exact canvas size. Pass format for others, and pixelRatio 2/3 for hi-res.

On the URL endpoint use ?format=webp&quality=80. Each format + quality is cached separately; identical requests return instantly and cost nothing.

Render one image synchronously. Results are cached by a content hash of the template + payload, so identical requests return instantly and cost nothing.

Request body

Payload value types

Each key is a layer name (from the Layers panel) or the element id — both work, the name wins if set. Each maps to one value object, depending on the element:

Text values take two optional per-render overrides: fontSize (px — overrides the auto-fit shrink) and color(any CSS colour). Long text always auto-shrinks to fit its box, so designs don't break.

{ "headline": { "text": "Big news", "fontSize": 64, "color": "#E11D48" } }

Example

curl -X POST https://render.pixeldrive.pro/v1/render \
  -H "Authorization: Bearer pdrv_..." \
  -H "Content-Type: application/json" \
  -d '{
    "templateId": "ee3a339f-2826-406c-a6e4-04b9a7b4d373",
    "payload": {
      "headline":  { "text": "500K followers" },
      "avatar":    { "src":  "https://example.com/pic.jpg" }
    },
    "pixelRatio": 1
  }'

Response

{
  "url": "https://render.pixeldrive.pro/img/t/<templateId>/<hash>.png",
  "fromCache": false,            // false | "redis" | "db"
  "fieldsHash": "a1b2c3d4e5f6g7h8",
  "renderMs": 620,
  "uploadBytes": 254714,
  "creditsRemaining": 9999
}

A URL-only render. Put it straight in an <img src>, an og:image tag, or an email — no code. It renders (or cache-hits) and 302-redirects to the immutable image URL.

Query params

Example

<img src="https://render.pixeldrive.pro/v1/render?templateId=ee3a339f-…&token=pdrv_…&headline=500K%20followers" />

# → 302 redirect → https://render.pixeldrive.pro/img/t/<templateId>/<hash>.png

Encode special characters

Field values must be URL-encoded. Characters like # & + % = are URL syntax — left raw, the value is silently truncated or split (a # drops everything after it before the request is even sent, so ?msg=A%23B arrives, but ?msg=A#B arrives as just A). Wrap each value in encodeURIComponent(), let curl -G --data-urlencode do it, or skip the URL and POST a JSON body (which needs no encoding).

# raw — BREAKS:  # truncates the URL, & splits it, % is a bad escape
?headline=Big news & 50% off! #sale

# encoded — works  (encodeURIComponent("Big news & 50% off! #sale"))
?headline=Big%20news%20%26%2050%25%20off!%20%23sale

The token sits in the URL, so treat these as semi-public: use a dedicated token you can revoke, or for truly public embeds, render server-side and embed the returned /img/… URL (which carries no token), optionally via /v1/sign.

Queue up to 1,000 renders against one template and get a batch id back immediately. Poll /v1/batches/:id or stream /stream for results. This is the right tool for generating hundreds of images — it never blocks a connection.

Request body

curl -X POST https://render.pixeldrive.pro/v1/render/batch \
  -H "Authorization: Bearer pdrv_..." \
  -H "Content-Type: application/json" \
  -d '{
    "templateId": "ee3a339f-...",
    "payloads": [
      { "name": { "text": "Alice" } },
      { "name": { "text": "Bob" } }
    ],
    "userId": "660e8400-..."
  }'

// 202 Accepted
{ "batchId": "770e8400-...", "count": 2, "jobIds": ["…","…"] }

Counts plus a per-job results array.

{
  "batchId": "770e8400-...",
  "total": 2, "queued": 0, "running": 1, "completed": 1, "failed": 0,
  "results": [
    { "id": "…", "status": "completed",
      "result": { "imageUrl": "https://render.pixeldrive.pro/img/…", "contentType": "image/png", "fromCache": false },
      "failedReason": null },
    { "id": "…", "status": "running", "result": null, "failedReason": null }
  ]
}

Server-Sent Events. Emits a progress event whenever counts change, then a done event with the full results. Closes automatically when finished (max 30 min).

event: progress
data: {"total":2,"completed":1,"failed":0,"running":1,"queued":0}

event: done
data: {"results":[ … ]}

Status of one job from a batch (ids come from the batch response's jobIds).

{
  "id": "job-uuid",
  "status": "completed",         // waiting | active | completed | failed | delayed
  "progress": 100,
  "result": { "imageUrl": "https://render.pixeldrive.pro/img/…", "contentType": "image/png" },
  "failedReason": null,
  "attemptsMade": 1
}

Public, content-addressed, cached for a year (immutable). These are the URLs returned by every render. Drop them straight into an <img>, an email, or a CDN.

GET https://render.pixeldrive.pro/img/t/ee3a339f-.../a1b2c3d4e5f6g7h8.png
→ 200, Content-Type: image/png, Cache-Control: public, max-age=31536000, immutable

Issue a short-lived signed URL for an image (for private/expiring links).

curl -X POST https://render.pixeldrive.pro/v1/sign \
  -H "Authorization: Bearer pdrv_..." \
  -d '{ "url": "https://render.pixeldrive.pro/img/t/…/….png", "expiresIn": 7200 }'

{ "url": "https://…signed…", "key": "t/…/….png", "expiresIn": 7200 }

A Model Context Protocol endpoint so AI assistants can browse, design, and render directly — 25 toolscovering the whole editor and renderer. Authenticate with the same Bearer API token. The two you’ll use most:

→ Full MCP guide: one-command install, one-click Cursor/VS Code buttons, the complete tool list

MCP-aware clients (Cursor, Claude Code, …)

{
  "mcpServers": {
    "pixeldrive": {
      "url": "https://pixeldrive.pro/mcp",
      "headers": { "Authorization": "Bearer pdrv_your_token" }
    }
  }
}

Claude Desktop (proxies the remote server over stdio)

{
  "mcpServers": {
    "pixeldrive": {
      "command": "npx",
      "args": [
        "-y", "mcp-remote", "https://pixeldrive.pro/mcp",
        "--header", "Authorization: Bearer pdrv_your_token"
      ]
    }
  }
}

Status codes

Manage tokens in API tokens · upgrade for more credits on Billing · see plans on Pricing.